'Ambiguous abbreviation' error - but not using any abbreviations - reghdfe

[Andrew Myers]

I have been running some higher dimensional fixed effects regressions using the reghdfe user-written module. I am using Stata 12/SE.
Some of my regressions run fine but some of them return an error.

For example when I run the following:

. eststo: quietly reghdfe ln_lead1_Q prop_rg prop_ir prop_aud prop_id cbt ln_size ln_ST ln_ST_prop_rg ln_ST_prop_ir ln_ST_prop_aud ln_ST_prop_id, absorb(i.group i.industry_year i.id_person) vce(cluster group)

I get this error:

ln_ST_prop_i ambiguous abbreviation
r(111);

end of do-file

r(111);

Hoping for some advice.

[Nick Cox]

Forum principle: When referring to user-written software you are asked to explain where it comes from. The fact of being user-written is only a start on that.

You are using reghdfe from somewhere.

You are also using eststo from somewhere. It is part of the estout package.

My guess: Inside some program, some list of variables is being truncated, possibly because of an out-dated local macro operation.

Possibility: You are not using the latest version of one of these programs.

Possibility: You have unearthed a bug.

Short of trying to replicate your problem with your data, the advice is to check for later versions and to use set trace on to see where the error occurs.

The users or authors of these programs may have further advice.

[Sergio Correia]

Nick's advice is--as usual--spot on.

An older SSC version of -reghdfe- was using macro operations that worked under Stata 13 but failed with Stata 12. Using the latest SSC version of -reghdfe- I ran the command and there were no reported problems, so my guess is that if you run

ado uninstall reghdfe
ssc install reghdfe

Then the problem will be solved (note: you need to uninstall first because the number of files changed between updates).

Best,
Sergio

[Andrew Myers]

I managed to solve the problem for myself. My solution was to rename the variables using fewer characters. Doing that eliminated the error.

Thank you for the advice still. I will also update reghdfe and will check later that that solves the problem even using my original variable names.

[Richard Williams]

Whenever I encounter what appears to be a bug, I always run -update all- (to make sure I have the most current version of Stata) and -adoupdate- (to make sure I have the most current version of user-written programs). If I am lucky the problem has been fixed. Conversely, I have had people write to me with problems in my programs, only to discover that I fixed the problem back in 2007.

With user-written routines, you also need to check whether the site you originally downloaded from actually has the most current version of the program. For example, a program that originally appeared in the Stata Journal may have a newer version of it available on SSC.